Dell Technologies (NYSE: DELL) announced the PowerEdge XE8812, a new server featuring Nvidia Vera Rubin NVL4 architecture, designed for high-performance computing and AI workloads. The announcement was made at ISC in Hamburg, Germany.

The fanless, direct liquid-cooled server scales to 144 GPUs per rack within the Dell PowerRack 9100, an OCP standards-based rack architecture. Compared to the prior generation using Nvidia GB200 NVL4, the XE8812 expands CPU cores from 144 to 176 and offers 50% more memory per socket and GPU memory. The server is based on the open ORv3 standard and supports 300kW+ power capacity.

Dell said the PowerEdge XE8812 will be globally available early next year.

The company cited several existing deployments of its Dell AI Factory with Nvidia platform. In the U.S., Dell, Nvidia and NERSC are building a supercomputer called Doudna at Lawrence Berkeley National Laboratory using PowerEdge XE8812 servers connected with Nvidia Quantum-X800 InfiniBand. In France, AI company InstaDeep is using the platform for its Kyber supercomputing cluster, which delivers approximately 0.5 exaFLOPs of FP16 performance. In the UK, the Wellcome Sanger Institute is using Dell PowerEdge XE-Series servers with Nvidia GPUs to process genomic data, producing one fully assembled genome every seven hours while managing over 100 petabytes of genetic data. In Australia, Monash University deployed the MAVERIC supercomputer using liquid-cooled Dell PowerRack systems with PowerEdge XE9712 servers and Nvidia GB200 NVL72 architecture.

Dell reported that more than 5,000 customers have deployed the Dell AI Factory globally, based on an April 2026 analysis of customer order data.